Acetyl-L-Carnitine Deficiency in Patients with Major Depressive Disorder

Acetyl-L-carnitine deficiency in patients with major depressive disorder

Carla Nascaa,1, Benedetta Bigioa,b, Francis S. Leec,d, Sarah P. Younge,f, Marin M. Kautzg, Ashly Albrightd, James Beasleyf, David S. Millingtone,f, Aleksander A. Mathéh, James H. Kocsisd, James W. Murroughg, Bruce S. McEwena,1, and Natalie Rasgona,i

aHarold and Margaret Milliken Hatch Laboratory of Neuroendocrinology, The Rockefeller University, New York, NY 10065; bBiostatistics and Experimental Research Design, Center for Clinical and Translational Science, The Rockefeller University, New York, NY 10065; cSackler Institute for Developmental Psychobiology, Weill Cornell Medical College, New York, NY 10065; dDepartment of Psychiatry, Weill Cornell Medical College, New York, NY 10065; eDivision of Medical Genetics, Department of Pediatrics, Duke University School of Medicine, Durham, NC 27710; fBiochemical Genetics Laboratory, Duke University Health System, Durham, NC 27710; gMood and Anxiety Disorders Program, Department of Psychiatry, Icahn School of Medicine at Mount Sinai, New York, NY 10029; hDepartment of Clinical Neuroscience, Karolinska Institutet, SE-171 77 Stockholm, Sweden; and iCenter for Neuroscience in Women’s Health, Stanford University, Palo Alto, CA 94305

Contributed by Bruce S. McEwen, June 15, 2018 (sent for review February 2, 2018; reviewed by Julio Licinio, Robert M. Post, and Charles L. Raison) The lack of biomarkers to identify target populations greatly limits neurotrophic factor (BDNF) and a critical regulator of synap- the promise of precision medicine for major depressive disorder tic glutamate release, the metabotropic glutamate receptor of (MDD), a primary cause of ill health and disability. The endoge- class-2, mGlu2 (10, 16–18). In several animal models, LAC nously produced molecule acetyl-L-carnitine (LAC) is critical for hip- supplementation has been shown to ameliorate glutamatergic pocampal function and several behavioral domains. In rodents dysfunction and associated neuronal atrophy in brain regions with depressive-like traits, LAC levels are markedly decreased such as the hippocampus and medial amygdala (13, 16, 19–21). and signal abnormal hippocampal glutamatergic function and den- LAC is an endogenous short-chain acetyl ester of free carnitine dritic plasticity. LAC supplementation induces rapid and lasting that crosses the blood–brain barrier (17, 22, 23). Notably, we antidepressant-like effects via epigenetic mechanisms of histone found that these animals that rapidly responded to LAC sup- acetylation. This mechanistic model led us to evaluate LAC levels in plementation (10–12, 19) show an endogenous decrease in LAC humans. We found that LAC levels, and not those of free carnitine, in plasma and in mood regulatory brain regions (i.e., the hip- MEDICAL SCIENCES were decreased in patients with MDD compared with age- and pocampus and prefrontal cortex) (13, 17). Furthermore, the sex-matched healthy controls in two independent study centers. deficiency in LAC was associated with insulin resistance (IR), Secondary exploratory analyses showed that the degree of LAC which was ameliorated by supplementation of LAC (19). deficiency reflected both the severity and age of onset of MDD. Therefore, the animal models provide a conceptual platform Moreover, these analyses showed that the decrease in LAC was larger in patients with a history of treatment-resistant depression that is consistent with a known role of glutamatergic dysfunction, altered trophic environment, and proinflammatory states in hu- (TRD), among whom childhood trauma and, specifically, a history – of emotional neglect and being female, predicted the decreased mans with depression (3, 21, 24 31). Here, using such a biological LAC. These findings suggest that LAC may serve as a candidate target- and mechanistically driven approach, we evaluated the role biomarker to help diagnose a clinical endophenotype of MDD of LAC in MDD in humans. characterized by decreased LAC, greater severity, and earlier onset as well as a history of childhood trauma in patients with TRD. Significance Together with studies in rodents, these translational findings sup- port further exploration of LAC as a therapeutic target that may Identifying biological targets in major depressive disorder (MDD) help to define individualized treatments in biologically based de- is a critical step for development of effective mechanism-based pression subtype consistent with the spirit of precision medicine. medications. The epigenetic agent acetyl-L-carnitine (LAC) has rapid and enduring antidepressant-like effects in LAC-deficient epigenetic | glutamate | treatment-resistant depression | rodents. Here, we found that LAC levels were decreased in pa- childhood trauma | mGlu2 tients with MDD versus age- and sex-matched healthy controls in two independent study centers. In subsequent exploratory ajor depressive disorder (MDD) is among the leading analyses, the degree of LAC deficiency reflected both the se- Mcauses of illness and disability worldwide (1, 2). MDD is a verity and age of onset of MDD. Furthermore, the lowest LAC severe and life-threatening disease, which is also associated with levels were found in patients with treatment-resistant de- other major illnesses, such as diabetes, cardiovascular disorders, pression, whereby history of emotional neglect and being fe- and Alzheimer’s disease (3, 4). A known risk factor for MDD is male predicted decreased LAC levels. These translational childhood trauma, which occurs at alarmingly high rates and has findings suggest that LAC may serve as a candidate biomarker been associated with poorer responses to available antidepressant to help the diagnosis of a clinical endophenotype of MDD. medications as well as with treatment-resistant depression (TRD) Author contributions: C.N., B.B., F.S.L., and B.S.M. designed research; F.S.L., S.P.Y., M.M.K., (5). The pathophysiology of MDD remains poorly understood, A.A., J.B., D.S.M., J.H.K., and J.W.M. performed research; C.N., B.B., and N.R. analyzed with a consequent lack of biological targets that can guide the data; C.N., A.A.M., B.S.M., and N.R. wrote the paper; and B.S.M. supervised research. development of diagnostics and improved therapeutics (6, 7). Reviewers: J.L., SUNY Upstate Medical University; R.M.P., Bipolar Collaborative Network; In rodent models, epigenetic agents such as histone deacety- and C.L.R., University of Wisconsin–Madison. lase inhibitors and the acetylating molecule acetyl-L-carnitine The authors declare no conflict of interest. A (LAC, Fig. 1 ) have been shown to promote rapid antidepres- Published under the PNAS license. – sant responses (8 15). Converging evidence from our group and 1To whom correspondence may be addressed. Email: [email protected] or mcewen@ others has shown that supplementation of LAC exerts rapid rockefeller.edu. antidepressant actions, at least in part, by acetylating histones to This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. regulate the expression of key genes important for synaptic 1073/pnas.1801609115/-/DCSupplemental. plasticity, including the proneurogenic molecule brain-derived

www.pnas.org/cgi/doi/10.1073/pnas.1801609115 PNAS Latest Articles | 1of6 Downloaded by guest on September 23, 2021 LAC Levels Within the MDD Group. Within the group of patients with MDD, no difference was observed in LAC concentrations in relation to use of psychotropic medications (SI Appendix, Fig. S3). Given the primary findings that showed a decrease in LAC levels in patients with MDD compared with age- and sex- matched HC in two independent study centers, we evaluated the contribution of clinical characteristics of MDD, such as depression severity and age of onset, on LAC levels by performing exploratory analyses. In subjects with mild MDD no correlation was observed between LAC levels and the severity of the disease using Hamilton Depression-Rating Scale (HDRS-17) (33). Among subjects with moderate to severe MDD, we observed a significant negative correlation between LAC concentrations and severity scores at HDRS-17 (33), whereby the higher the severity the lower the concentrations of LAC (P = 0.04, r = 0.35) (SI Appendix,Fig.S4). This relationship remained significant upon multiple regression analysis controlling for number of past episodes (t = −2.13, P = 0.04) and length of current episode (t = −2.57, P = 0.017) as well as controlling for sex (P < 0.0001) and age (P = 0.0001). Furthermore, prediction model analysis showed that LAC predicted HDRS-17 severity scores among subjects with moderate to severe MDD (P = 0.04, r = 0.35, SI Appendix, Fig. S5). Indeed, the HDRS-17 severity scores statistically inferred from LAC measures were consistent with the rater-administered HDRS-17 severity scores (SI Appendix,Fig.S5). Pearson correlation analysis also showed a positive correlation between LAC concentrations and age of onset of depression (P = 0.04, r = 0.32); that is, earlier age of onset correlated with lower concentrations of Fig. 1. Decreased Acetyl-L-carnitine (LAC) Levels in patients with MDD com- SI Appendix pared with HC. (A) Schematic model featuring our innovative framework: The LAC ( ,Fig.S6). endogenously produced molecule acetyl-L-carnitine (LAC) is critical for hippocampal function and several behavioral domains. In rodents with depressive- LAC Levels and Treatment-Resistant Depression: Role of Childhood like traits, LAC levels are markedly decreased and accompanied by abnormal Trauma. Driven by the findings above showing an LAC de- hippocampal glutamatergic function, decreased expression of the neuro- ficiency in patients with MDD in two study centers, we evaluated trophic factor BDNF, and dendritic plasticity as well as by systemic in- LAC levels across HC subjects and patients with MDD with or flammation, including insulin resistance. LAC supplementation rescues those without history of TRD. Consistent with lower LAC in patients deficits and induces rapid and lasting epigenetic antidepressant-like effects via acetylation of histones. (B and C)PlasmaLAC(B) and free-carnitine (C)con- with MDD that were also characterized by greater severity and centrations in HC and in patients with MDD in acute depressive episodes earlier onset of the illness, we found that the decrease in LAC during study participation as assessed by ultraperformance liquid chromatography–electrospray–tandem mass spectrometry (UPLC-MS/MS). See also SI Appendix, Figs. S1 and S3 *Significant comparisons with HC. ***P < 0.001 in Student’s two-tailed t tests (α = 0.05). Dashed bars indicate group mean.

Results LAC Levels Differ Between Healthy Controls and Patients with MDD. In a sample of 116 participants, no difference was observed between healthy controls (HC) (n = 45) and patients with MDD (n = 71) with respect to demographic characteristics, including age and sex (SI Appendix, Tables S1 and S2). All patients were in an acute depressive episode during study participation. LAC levels (Fig. 1A) in plasma, as the most accessible bio- specimen in a clinical setting, were measured by UPLC-MS/MS (ultraperformance liquid chromatography–electrospray–tandem mass spectrometry) and ESI-MS/MS (stable isotope dilution electrospray–tandem mass spectrometry) as previously described (10, 32). LAC differed significantly between HC and patients with MDD with mean concentrations lower in the MDD group (Fig. 1B, P < 0.0001, effect size = 0.8; HC: 8.3 μmol/L ± 0.4, MDD: 6.1 μmol/L ± 0.3; SI Appendix, Fig. S1). No significant difference was observed in free-carnitine concentrations between HC and patients with MDD (Fig. 1C). The association between MDD and LAC held when stratifying by sex (SI Appendix, Fig. Fig. 2. LAC levels differ between HC and MDD groups in two independent study centers. Plasma LAC levels were similarly decreased in patients with S2). Furthermore, LAC levels were similarly decreased in pa- MDD compared with age- and sex-matched HC in both study centers (C, tients with MDD compared with age- and sex-matched HC in Cornell; S, Sinai) as assessed by UPLC-MS/MS. *Significant comparisons with both independent study centers [Weill Cornell (C) and Mount HC. **P < 0.01 in Student’s two-tailed t tests (α = 0.05). Dashed bars indicate Sinai School of Medicine (S)] (Fig. 2). group mean.

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1801609115 Nasca et al. Downloaded by guest on September 23, 2021 was greater in subjects with MDD and a history of TRD, F(2, 53) = LAC levels are markedly decreased and signal abnormal brain 4.3, P = 0.01 (Fig. 3 and SI Appendix,Fig.S7). functions as well as metabolic dysregulation (SI Appendix, Fig. As childhood trauma has been associated with depression S8). Furthermore, hypotheses-generating exploratory analyses severity and treatment resistance, we conducted exploratory reported herein reveal that (i) the degree of LAC deficiency analyses to evaluate the contribution of childhood trauma on reflected both the severity and age of onset of MDD, and (ii) the LAC levels. First, we found that the reported rates of childhood LAC deficiency was associated with a history of childhood trauma assessed by the Childhood Trauma Questionnaire (CTQ) trauma in patients with treatment-resistant depression (SI Ap- (34) differed from HC and patients with MDD (Fig. 4). Within pendix, Fig. S8). These findings compel further research on the the group of patients with MDD, we found that rates of emo- potential role of LAC as a candidate biomarker that together tional abuse, physical neglect, and emotional neglect were sig- with clinical characteristics can aid the diagnosis and identifica- nificantly higher either in patients with TRD or in patients with tion of a clinical endophenotype of MDD. Furthermore, LAC MDD non-TRD compared with HC, while physical and sexual deficiency may represent an innovative biological therapeutic abuse were significantly higher only in patients with TRD versus target in treatment of depression. HC (Fig. 4 A–E): emotional abuse, F(2, 52) = 17.08, P < 0.0001; The decreased levels of LAC in patients with MDD is partic- physical abuse, F(2, 52) = 6.88, P = 0.0023; sexual abuse, F(2, 52) = ularly important because LAC is an essential molecule for sys- 2.35, P = 0.1; emotional neglect, F(2, 52) = 13.75, P < 0.0001; temic and neural functions (13, 17, 22). LAC plays a central role physical neglect, F(2, 52) = 6.26, P = 0.0038. in the transport of fatty acids into the mitochondria for beta ox- Furthermore, multiple regression analyses showed that LAC idation to sustain energy metabolism in the brain and the rest of levels were predicted by a history of childhood trauma in patients the body. LAC also facilitates elimination of oxidative products, with TRD; that is, emotional neglect by sex interaction predicted provides acetyl groups to regulate expression of neurotrophins LAC levels (P = 0.04, r = 0.66, Fig. 4F). Specifically, models and glutamate genes that contain spontaneous glutamate release, and protects from excitotoxicity, therefore interacting with stratified by sex showed emotional neglect as a predictor of LAC mechanisms that contribute to the pathophysiology of MDD (10, concentrations only in women with TRD (P = 0.02, r = 0.89, Fig. 13, 17, 21). In rodent models with depressive-like traits, LAC 4G), but not in men (P = 0.59, r = 0.40, Fig. 4H). No interaction levels are markedly decreased and accompanied by hippocampal of LAC with emotional neglect was observed in women HC (P = P = glutamatergic dysfunction as well as abnormal dendritic plasticity 0.77) or in women with MDD non-TRD ( 0.84). in the hippocampus, among other brain regions (13, 16, 20). LAC supplementation ameliorates these deficits. The specificity of

Discussion MEDICAL SCIENCES changes in LAC and lack of changes in free carnitine suggest that We report that a deficiency of the epigenetic agent acetyl-L- the relationship between LAC and MDD is independent of po- carnitine (LAC) occurs in humans who have major depressive tential dietary changes (17, 35). Of interest is also the previously disorder. These translational findings are an outgrowth of a reported positive correlation between peripheral and CNS LAC mechanistic model in rodents with depressive-like traits, wherein concentrations (17). Of note, LAC levels were similarly decreased in patients with MDD compared with age- and sex-matched HC in both study sites, providing an initial replication of our results. Furthermore, the use of two different methods to assess LAC in the same 116 samples supports the reliability of our results. Fu- ture studies will also be needed to assess whether decreased LAC levels in patients with MDD are sensitive to unhealthy lifestyle choices, such as physical inactivity, poor dietary habits, and lack of adequate sleep. Furthermore, LAC appears to act as a state-dependent marker, given that all patients were in an acute depressive episode at the time of study participation and that the presence of medications did not influence LAC levels. However, the cross- sectional design of our study does not allow establishing whether the decrease in LAC levels in patients with MDD may also be a trait biomarker. Further studies may help in elucidating trait- dependent LAC levels. In addition, given the high comorbidity of MDD with other psychiatric disorders, future studies with larger cohorts will be needed to investigate whether decreased LAC levels may represent a specific signature for MDD or is a general marker of affective disorders. Moreover, the finding that the decrease in LAC levels in patients with MDD is independent of psychotropic drug treatment raises the possibility that increasing LAC levels may be needed to induce antidepressant effects. LAC levels were significantly correlated with depression severity and with age of onset of MDD. Although these were exploratory analyses, the strength of such correlations remained aftercontrollingforsex,numberofdepressiveepisodes,anddu- ration of the current episode and was independent of use of psychotropic medications. Consistent with these results is the Fig. 3. The LAC deficiency is greater in patients with MDD and history of observation of decreased LAC in patients with history of treatment-resistant depression (TRD). Plasma LAC concentrations across HC, patients with MDD without history of TRD (MDD non-TRD), and patients treatment-resistant depression (TRD). The greater decrease in = = LAC in more severe forms of MDD and in patients with TRD is with MDD and with history of TRD for study center S. F(2, 53) 4.3, P 0.01. “ ” *Significant comparisons with HC. *P < 0.05, **P < 0.01 in Student’stwo- akin to a kindling-like progression of MDD in that earlier age at tailed t tests (α = 0.05). Dashed bars indicate group mean. SI Appendix, Fig. onset and/or the presence of early life adversity, such as childhood S7 also shows a consistency of this stepwise in study center C. trauma, conveys liability to more severe and treatment-resistant

Nasca et al. PNAS Latest Articles | 3of6 Downloaded by guest on September 23, 2021 Fig. 4. LAC as potential moderator of sex-specific effects of childhood trauma in patients with MDD with or without history of treatment-resistant depression (TRD). (A–H) History of childhood trauma as assessed by the Childhood Trauma Questionnaire (CTQ) with individual areas, including psychical abuse (A), sexual abuse (B), emotional abuse (C), physical neglect (D), and emotional neglect (E)inHC(n = 19), in patients with MDD non-TRD (n = 16), and in patients with TRD (n = 18) that reported childhood trauma as assessed using by the CTQ at study center S. Data are presented as mean ± SEM. (F) Multiple regression analysis of LAC by emotional neglect and sex in patients with TRD. In the x axis: LAC concentrations as predicted by the model; in the y axis: LAC concentrations as measured in patients with TRD. (G and H) Models stratified by sex in women (G) and men (H) with TRD. *Significant comparisons with HC; #Significant comparison with MDD non-TRD. *P < 0.05, **P < 0.01, ***P < 0.001 in Student’stwo-tailedt tests (α = 0.05) or using multiple regression analysis for the predictive model.

course of illness (35). Our exploratory analyses also revealed that a the same animal models (10–12, 14, 18, 19). To the best of our history of emotional neglect predicted LAC levels in women with knowledge, there is no clinical study to date that tested the efficacy TRD, suggesting that LAC may moderate sex-specific effects of of LAC in patients with MDD. Previous studies showed that LAC childhood adversities in patients with TRD. This finding is con- treatment is well tolerated and effective in treatment of depressive sistent with previous epidemiological studies showing sex-specific symptoms, but these studies mainly focused on limited cohorts of correlations between childhood trauma and depressive symptoms elderly patients with dysthymia or fibromyalgia (13, 39, 40). The (36) and that the consequences of neglect differ substantially from conceptual framework that we pursue suggests an LAC deficiency those of other traumas (5, 36). Together with findings from pre- as a potential therapeutic target in the pathophysiology of MDD vious clinical studies that a history of childhood trauma impairs toward the development of more effective precision medicine responses to available antidepressant medications (37, 38), the approaches tailored to specific clinical biobehavioral phenotypes. current findings of LAC deficiency suggest a clinical endopheno- Indeed, LAC levels, together with clinical characteristics (i.e., type characterized by greater severity, earlier age of onset MDD, a MDD severity scores) and developmental history (i.e., age of history of treatment-resistant course of the illness, and childhood MDD onset and childhood trauma), may serve to identify specific trauma. Of note, and as a limitation, we had CTQ information clinical phenotypes of depression. Such phenotypes may be more only in patients from study center S; therefore, it would be im- likely to benefit from a biologically based treatment with LAC portant to ascertain the role of childhood trauma on the LAC supplementation or augmentation. Within this framework, it is deficiency with larger cohorts. important to emphasize that clinical trials of acute treatment with These current translational findings are an extension of pre- LAC are needed to validate the current postulate. Furthermore, it clinical research that shows that in rodents characterized by an will be important to test whether LAC supplementation has pre- LAC decrease and depressive-like phenotypes, supplementation ventive effects given its long-lasting antidepressant action in ani- with LAC leads to antidepressant responses seen after 3 d that also mals and evidence of induction of resilience. Based upon our last for 14 d after drug withdrawal. Responses to standard antide- earlier reported association of decreased LAC with insulin re- pressant medications require repeated weeks of administration in sistance (IR) in animals with depressive-like behaviors (19), it will

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1801609115 Nasca et al. Downloaded by guest on September 23, 2021 also be important to investigate the association of LAC deficiency LAC Measures by Ultraperformance Liquid Chromatography–Electrospray– with insulin resistance. Tandem Mass Spectrometry. LAC and free carnitine in plasma were ana- – In conclusion, the current findings of LAC deficiency in MDD lyzed using ultraperformance liquid chromatography tandem mass spectrom- suggest a possible endophenotype of depression, characterized by etry (UPLC-MS/MS) with electrospray ionization in positive ion mode on an Xevo-TQD or a TQD tandem mass spectrometer equipped with Acquity UPLC history of childhood trauma, greater depression severity, and system (Waters Corp.). For the determination of free L-carnitine and LAC, 2 2 earlier age at onset. Future prospective, placebo-controlled studies plasma samples were spiked with [ H3]-free carnitine and acetyl–[ H3]-carnitine will be needed to address some of the limitations inherent in our internal standards. The total concentration of carnitine (sum of free and acyl- cross-sectional cohorts. Further study with larger cohorts is also ated carnitine) was determined in a second aliquot of each sample mixed with 2 needed to better understand the role of LAC in clinically distinct [ H3]-free carnitine internal standard. These second aliquots were subjected to populations of patients with depressive disorders. base hydrolysis of the acylcarnitine species by incubation with 1 mol/L KOH at 65 °C for 15 min, followed by neutralization with 1 mol/L HCl. Protein was Methods precipitated in all aliquots using 0.1% formic acid in acetonitrile and removed by centrifugation. The sample extracts were dried and reconstituted in 0.1% The Rockefeller University Institutional Review Board and the respective formic acid and 7.5 mmol/L ammonium formate in 18:82 (vol/vol) acetonitrile: Institutional Review Boards of the collaborating Institutions (Weill Cornell deionized water. L-Carnitine and LAC were separated on an Acquity BEH HILIC, Medicine, Icahn School of Medicine at Mount Sinai, and Duke University) 2.1 mm × 100 mm, 1.7-μm column (Waters Corp.) with gradient elution using approved the current study in its entirety. 0.1% formic acid and 7.5 mmol/L ammonium formate in acetonitrile:deionized water as the mobile phase and detected using selected reaction monitoring. Participants. Following an initial phone screen, potential participants were The ratios of signal intensities for the transitions m/z 162 > 103 (free carnitine) evaluated in person to determine study eligibility. All participants de- 2 and 165 > 103 ([ H3]-carnitine) and m/z 204 > 85 (acetylcarnitine) and 207 > 85 termined to be eligible to join the study provided written informed consent 2 (acetyl–[ H3]-carnitine) were converted to a concentration by means of a cali- before study enrollment. Study participants, ranging between 20 and 70 y bration curve. Materials: L-carnitine.HCl and LAC hydrochloride (Sigma Chemical old, were recruited at two independent sites, the Affective Disorders Re- 2 2 Co.); H3-L-carnitine.HCl and acetyl– H3-L-carnitine.HCl (Cambridge Isotope search Program at Weill Cornell Medicine and the Mood and Anxiety Dis- Laboratories, Inc.); all other reagents, solvents, and solvent additives were orders Program at the Icahn School of Medicine at Mount Sinai. At both study purchased from Sigma Chemical Co. or VWR. All groups were evenly divided sites, study clinicians or trained coordinators conducted the Structured between the experimental plates to account for any interplate variability. Clinical Interview for DSM-IV (SCID) or Mini International Neuropsychiatric Interview (MINI) to confirm MDD diagnosis and rule out exclusionary LAC Measures by Stable Isotope Dilution Electrospray–Tandem Mass comorbid conditions. Spectrometry. Deuterated internal standards were obtained from Cam- Inclusion and exclusion criteria were similar at both recruitment sites. bridge Isotope Laboratories, Inc. and from Sigma Chemical Co. Three Molars Inclusion criteria included a primary diagnosis of MDD in a current major methanolic HCl was obtained from Sigma Chemical Co. General reagents and MEDICAL SCIENCES depressive episode. Inclusion criteria in the TRD group also included having at solvents were obtained from VWR. Plasma acetylcarnitine was analyzed as least moderate severity and a history of nonresponse to at least two ther- methyl ester by a semiquantitative method using stable isotope dilution elec- apeutic trials of an antidepressant according to the Antidepressant Treat- trospray–tandem mass spectrometry (ESI-MS/MS). Plasma was mixed in an in-

ment History Form (ATHF) or Antidepressant Treatment Record (ATR) during ternal standard mixture containing d3-acetylcarnitine, d3-propionylcarnitine, their lifetime. Exclusion criteria included a presence of neurologic or other d3-butylcarnitine, d3-octanoylcarnitine, and d3-palmitoylcarnitine. Protein was physical illness, such as diabetes, alcohol or substance abuse in the last 6 mo, precipitated by the addition of methanol and removed by centrifugation. An or an unstable medical illness. Current medication use was assessed at aliquot of the supernatant liquid was dried under nitrogen and methylated by screening for all study participants. Participants were free of current sub- incubation with 3 M HCl in methanol at 50 °C for 15 min. The derivatized ex-

stances of abuse as determined by a urine toxicology test at the time of tract was dried under nitrogen, reconstituted in methanol:dH2O 85:15 (vol/vol), screening. HC were free of lifetime psychiatric illness and significant medical and analyzed directly by flow injection–MS/MS on a TQD tandem mass spec- conditions. Participants were free of active infections and systemic illness as trometercoupledwithanAcquityUPLC system (Waters Corporation). Ace- confirmed by medical history at the time of study evaluation. Blood samples tylcarnitine was detected using a precursor ion scan of m/z 99, with a scan range were obtained via antecubital venous collection using standard techniques of m/z 200–500. Concentrations were determined from the ratio of ion in- and were drawn after a period of fasting (>6 h). Participants were asked not tensities of acetylcarnitine species to its specified deuterated internal standard, to exercise for >6 h before blood draw. multiplied by the concentration of the standard. All groups were evenly divided between the experimental plates to account for any interplate variability. Clinical and Psychiatric Assessment. Clinical assessment consisted of a physical examination, including measures of height, weight, and body mass index Statistical Analysis. Statistical analyses were conducted using JMP Software (BMI). Other data collected included current medication use and history of from SAS (Statistical Analysis Systems Institute). Two-tailed t tests and χ2 failed antidepressant trials. Demographic information, including sex, was analyses were used to compare, respectively, continuous and categorical also recorded from the participants (SI Appendix,TablesS1andS2). The demographic and clinical characteristics between HC and MDD subjects. psychiatric examination at screening included SCID or MINI to confirm Between- and within-group differences in patient plasma LAC and free- MDD diagnosis, and trained raters administered the structured depression- carnitine concentrations were compared using t tests. Within-group Pear- rating scales: 17-item Hamilton Depression-Rating Scale (HDRS-17) and the son correlations were conducted to examine the relationship between LAC Montgomery–Asberg Depression Rating Scale (MADRS). Among partici- concentrations and depression severity or age of onset. Multiple regression pants, two were identified as having eating disorders and one as having analysis was used to control for other clinical characteristics. A one-way Crohn’s disease and, therefore, were excluded from the analyses. Cutoff ANOVA followed by post hoc Student’s t tests was used to examine LAC scores of 19 and 34 were used to stratify for depression severity at the levels upon use of psychotropic medications as well as across HC and subjects HDRS-17 and MADRS, respectively (33). As a result, among the 71 patients with MDD with or without history of TRD. Predictive models were inferred with MDD, 28 patients had moderate depression (HDRS-17 < 19, MADRS < using multiple regression analysis to assess the ability of LAC and CTQ areas 34) and 43 patients had severe depression at the time of study evaluation to predict the dependent variables, HDRS-17 scores, and LAC concentrations, ± (HDRS-17 ≥ 19, MADRS ≥ 34). With regard to medication use, 53 patients respectively. Significance was set at 0.05, and data are presented as mean were free of antidepressant medications at the time of study participation SD, unless otherwise specified. and 18 patients were on psychotropic medications. A subgroup of 18 patients with MDD recruited at Icahn School of Medicine at Mount Sinai ACKNOWLEDGMENTS. This work was supported by a grant from the reported history of TRD defined by at least two failed antidepressant trials Robertson Foundation (to C.N.) and, partly, by a grant from the Hope for (41, 42). All participants (i.e., HC, MDD non-TRD, and TRD) at Icahn School Depression Research Foundation (HDRF) (to B.S.M.). This work was also of Medicine at Mount Sinai completed the Childhood Trauma Question- funded in part by the Pritzker Neuropsychiatric Disorders Research Consor- tium, which is supported by the Pritzker Neuropsychiatric Disorders Research naire (CTQ) (43) to assess for childhood traumatic experiences in five Fund L.L.C. A shared intellectual property agreement exists between this specific areas: physical, sexual, and emotional abuse and physical and philanthropic fund and the University of Michigan, Stanford University, the emotional neglect. Some information about subjects from the Icahn Weill Medical College of Cornell University, the University of California at School of Medicine at Mount Sinai included in the current study was Irvine, and the Hudson Alpha Institute for Biotechnology to encourage the previously reported (44). development of appropriate findings for research and clinical applications.

Nasca et al. PNAS Latest Articles | 5of6 Downloaded by guest on September 23, 2021 1. Kessler RC, Bromet EJ (2013) The epidemiology of depression across cultures. Annu 25. Lin KW, Wroolie TE, Robakis T, Rasgon NL (2015) Adjuvant pioglitazone for un- Rev Public Health 34:119–138. remitted depression: Clinical correlates of treatment response. Psychiatry Res 230: 2. Friedrich MJ (2017) Depression is the leading cause of disability around the world. 846–852. JAMA 317:1517. 26. Rasgon NL, et al. (2010) Rosiglitazone add-on in treatment of depressed patients with 3. Rasgon NL, McEwen BS (2016) Insulin resistance: A missing link no more. Mol insulin resistance: A pilot study. Sci World J 10:321–328. – Psychiatry 21:1648 1652. 27. Duman RS, Li N (2012) A neurotrophic hypothesis of depression: Role of synapto- 4. McIntyre RS, et al. (2009) Metabolic syndrome and major depressive disorder: Co- genesis in the actions of NMDA receptor antagonists. Philos Trans R Soc Lond B Biol – occurrence and pathophysiologic overlap. Curr Diab Rep 9:51 59. Sci 367:2475–2484. 5. Nemeroff CB (2016) Paradise lost: The neurobiological and clinical consequences of 28. Zarate CA, Jr, Machado-Vieira R (2017) Ketamine: Translating mechanistic discoveries child abuse and neglect. Neuron 89:892–909. into the next generation of glutamate modulators for mood disorders. Mol Psychiatry 6. Schmidt HD, Shelton RC, Duman RS (2011) Functional biomarkers of depression: Di- 22:324–327. agnosis, treatment, and pathophysiology. Neuropsychopharmacology 36:2375–2394. 29. Monteggia LM, Zarate C, Jr (2015) Antidepressant actions of ketamine: From mo- 7. Gururajan A, Clarke G, Dinan TG, Cryan JF (2016) Molecular biomarkers of depression. lecular mechanisms to clinical practice. Curr Opin Neurobiol 30:139–143. Neurosci Biobehav Rev 64:101–133. 30. Raison CL, Miller AH (2011) Is depression an inflammatory disorder? Curr Psychiatry 8. Nestler EJ (2014) Epigenetic mechanisms of depression. JAMA Psychiatry 71:454–456. – 9. Tsankova NM, et al. (2006) Sustained hippocampal chromatin regulation in a mouse Rep 13:467 475. model of depression and antidepressant action. Nat Neurosci 9:519–525. 31. Miller AH, Raison CL (2016) The role of inflammation in depression: From evolu- tionary imperative to modern treatment target. Nat Rev Immunol 16:22–34. 10. Nasca C, et al. (2013) L-acetylcarnitine causes rapid antidepressant effects through the epigenetic induction of mGlu2 receptors. Proc Natl Acad Sci USA 110:4804–4809. 32. Lu Y, et al. (2016) Acetylcarnitine is a candidate diagnostic and prognostic biomarker – 11. Russo SJ, Charney DS (2013) Next generation antidepressants. Proc Natl Acad Sci USA of hepatocellular carcinoma. Cancer Res 76:2912 2920. 110:4441–4442. 33. (2018) Inventory of Depressive Symptomatology (IDS) and Quick Inventory of De- 12. Flight MH (2013) Antidepressant epigenetic action. Nat Rev Neurosci 14:226. pressive Symptomatology (QIDS). Available at www.ids-qids.org/interpretation.html. 13. Wang SM, et al. (2014) A review of current evidence for acetyl-L-carnitine in the Accessed June 15, 2018. treatment of depression. J Psychiatr Res 53:30–37. 34. Bernstein DP, et al. (1994) Initial reliability and validity of a new retrospective mea- 14. Wang W, et al. (2015) Rapid-acting antidepressant-like effects of acetyl-L-carnitine sure of child abuse and neglect. Am J Psychiatry 151:1132–1136. mediated by PI3K/AKT/BDNF/VGF signaling pathway in mice. Neuroscience 285: 35. van Vlies N, et al. (2005) Characterization of carnitine and fatty acid metabolism in 281–291. the long-chain acyl-CoA dehydrogenase-deficient mouse. Biochem J 387:185–193. 15. Madiraju P, Pande SV, Prentki M, Madiraju SR (2009) Mitochondrial acetylcarnitine 36. Garcia M, et al. (2016) Sex differences in the effect of childhood trauma on the clinical provides acetyl groups for nuclear histone acetylation. Epigenetics 4:399–403. expression of early psychosis. Compr Psychiatry 68:86–96. 16. Lau T, Bigio B, Zelli D, McEwen BS, Nasca C (2017) Stress-induced structural plasticity 37. Williams LM, Debattista C, Duchemin AM, Schatzberg AF, Nemeroff CB (2016) of medial amygdala stellate neurons and rapid prevention by a candidate antide- Childhood trauma predicts antidepressant response in adults with major depression: – pressant. Mol Psychiatry 22:227 234. Data from the randomized international study to predict optimized treatment for – 17. Pettegrew JW, Levine J, McClure RJ (2000) Acetyl-L-carnitine physical chemical, met- depression. Transl Psychiatry 6:e799. ’ abolic, and therapeutic properties: Relevance for its mode of action in Alzheimer s 38. Kapczinski F, et al. (2008) Allostatic load in bipolar disorder: Implications for patho- disease and geriatric depression. Mol Psychiatry 5:616–632. physiology and treatment. Neurosci Biobehav Rev 32:675–692. 18. Cuccurazzu B, et al. (2013) Upregulation of mGlu2 receptors via NF-κB p65 acetylation 39. Meister R, et al. (2016) Comparative safety of pharmacologic treatments for persistent is involved in the proneurogenic and antidepressant effects of acetyl-L-carnitine. depressive disorder: A systematic review and network meta-analysis. PLoS One 11: Neuropsychopharmacology 38:2220–2230. e0153380. 19. Bigio B, et al. (2016) Epigenetics and energetics in ventral hippocampus mediate rapid 40. Veronese N, et al. (2018) Acetyl-L-carnitine supplementation and the treatment of antidepressant action: Implications for treatment resistance. Proc Natl Acad Sci USA depressive symptoms: A systematic review and meta-analysis. Psychosom Med 80: 113:7906–7911. – 20. Nasca C, et al. (2017) Role of the astroglial glutamate exchanger xCT in ventral hip- 154 159. pocampus in resilience to stress. Neuron 96:402–413.e5. 41. Souery D, Papakostas GI, Trivedi MH (2006) Treatment-resistant depression. J Clin – 21. McEwen BS, et al. (2015) Mechanisms of stress in the brain. Nat Neurosci 18: Psychiatry 67:16 22. 1353–1363. 42. Russell JM, et al. (2004) The cost consequences of treatment-resistant depression. 22. Fritz IB, McEwen B (1959) Effects of carnitine on fatty-acid oxidation by muscle. J Clin Psychiatry 65:341–347. Science 129:334–335. 43. Bernstein DP, et al. (2003) Development and validation of a brief screening version of – 23. Scafidi S, et al. (2010) Metabolism of acetyl-L-carnitine for energy and neurotrans- the Childhood Trauma Questionnaire. Child Abuse Negl 27:169 190. mitter synthesis in the immature rat brain. J Neurochem 114:820–831. 44. Kiraly DD, et al. (2017) Altered peripheral immune profiles in treatment-resistant 24. Mitani H, et al. (2006) Correlation between plasma levels of glutamate, alanine and serine depression: Response to ketamine and prediction of treatment outcome. Transl with severity of depression. Prog Neuropsychopharmacol Biol Psychiatry 30:1155–1158. Psychiatry 7:e1065.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1801609115 Nasca et al. Downloaded by guest on September 23, 2021